Centralized and Decentralized Automation Concepts - Part 2
An increasing complexity in the economic world also requires increasingly complex systems and procedures. Automation concepts play an increasingly significant role here, whether in agriculture, in the automobile sector, or only in smaller companies. At the same time, the decision for a centralized or a decentralized automation concept depends on different factors like the necessity of a precise and perfectly timed control. Therefore each of these concepts involves special advantages and disadvantages and can be applied on special fields of activity.
What are Centralized and Decentralized Automation Concepts?
It should be clarified what is meant by both concepts to provide clarity. The decentralized concept means in detail "distributed intelligence" because many distributed controls take over a big part of the task processing here. Even though such concepts usually permit higher speeds and allow modular machine concepts. For this reason, decentralized automation concepts are often more suitable for machine automation with many drive functions. Just to give one example, POWERLINK is such a system which can be also suitable for centralized solutions. In plain language, this concept means that the main part of the tasks goes to one or a few centralized controls and are being performed there, and only small working parts are being outsourced and assigned to network nodes. The big advantage of such centralized concepts lies in the application to the process industry because plants require a very high degree of automation and can be operated safe and secure by only a few required persons. These are only some nuances in definitions to differentiate between both concepts. Which concept is recommended specifically for which area depends on the required application speed.
Essential Criterion: The Application Speed
The time control of such processes and the determination of this control primarily decide whether a centralized or a decentralized system should and can be used. You can talk about a so-called "degree of hardness" of real-time because principally you can say that with an increase of this degree the control moves away from a centralized structure and to a decentralized structure.
The boundary of both concepts is not clearly drawn and intermingles, because a centralized automation concept also includes decentralized components and vice versa, a decentralized concept also has some centralized features. The differences between the concepts are in the scope of the tasks which have been allocated for control. The decentralized control outsources whole tasks, while in the decentralized structure only smaller "working parts" are mentioned. A decentralized concept consequently relieves the centralized control through this outsourcing to intelligent components. With this, a higher and better communication link of these components with the centralized system it is necessary.
The decision regarding the concepts also depends on which network structures can fulfil the timing demands best and with the least effort. For centralized systems for example, it can be assumed that with high data traffic there is only a moderate demand on the total computer performance and the transmission speed. The real-time therefore represents only a "soft" component, i.e. the period of time between the triggering and the occurrence of the event has a duration between one hundredth and one thousandth second. The requirements are therefore more likely suitable for a centralized automation concept and can be well applied in the process industry. Contrary to this, in the process industry "hard" real-times are also often required. The duration is in the µ-second range, which is why only decentralized concepts are suitable and are being applied for these processes. The advantage of distributed intelligence comes take effect with such complex systems.
Modular Machine Concepts - which Concepts can be applied better?
As processes require a higher flexibility and higher speed at the same time, and as these features only exist partly in centralized systems, this resulted in the development of modular machine concepts. These expand the scope of functions in general by connecting independent components without the need to modify the machine core.
With these, the drive control has to perform a variety of tasks which comprise for example regulation of the speed or calculation of the axis acceleration per phase. These calculations take place at exactly two different places: In the drive itself, which implies that corresponding processing units also exist there, or in the centralized control itself. It is obvious here that a centralized calculation requires a wide bandwidth of the bus and an immense processing power of the centralized control is required. It is also obvious that the data volumes can take on an extreme extent, which has a negative effect on the speed. The calculations would only be done by the master to send the results afterwards to the slave that has to give continuous feedback over the bus about its present condition. This fact results in an increasing susceptibility of the application function.
Having the computing power take place external is nevertheless hardly practicable, which is why the advantages of a decentralized concept are obvious in this application. Faster computing and reaction times exist, as well as a relief of the master is given. This in turn relieves the data traffic of the bus. An easier, modular expandability of the machinery through decentralized concepts is also given, as well as a safe application. To generate all these advantages, a correspondingly good communication connection must exist.